How can I optimize the performance of a Python script that processes large CSV files with millions of rows, and which libraries (such as Pandas, Dask) are most suitable for this task?